This document describes the configuration of StrathE2E2 for the Norwegian basin and its parameterisation to enable stationary state fitting for 2010-2070. These represent contrasting periods of environmental conditions.
Volumetric and seabed habitat data define the physical configuration of the system. We regard these as being fixed in time. Similarly, we regard the physiological parameters of the ecology model as being fixed in time. Some of these are set from external data. The remainder are fitted, as detailed here. Changes in the model performance between the different time periods therefore stem from the hydrodynamic, hydro-chemical and fishery driving data. These are detailed in the ecological drivers and fishing fleet sections.
Department of Mathematics and Statistics, University of Strathclyde, Glasgow, UK.
E-mail: m.heath@strath.ac.uk
The code written to support this parameterisation is available on github.
WARNING: This is a working document, subject to update and revision.
The model splits the domain into three zones, inshore/shallow, offshore/shallow, and offshore/deep (Figure 1). The inshore/shallow zone covers waters shallower than 60 m . The offshore zone covers the remaining area of the model domain (Figure 2). The offshore zone is divided further into a shallow and deep layer. The shallow layer represents water from the surface to 60 m depth, and shares a boundary with the inshore shallow zone. The offshore/deep zone covers the same area as the offshore/shallow zone, but represents water between 60 m and 600 m deep. There is a second internal boundary between the two offshore zones.
The seafloor of the model domain is represented by 5 habitat types. There are three sediment classes for the inshore zone – fine (muddy, 1), medium (sandy, 2) and coarse (gravel, 3). The fourth class (rock, 0) represents an absence of soft sediment. The rock class has different geochemical properties and supports the kelp forests in the model food web. In the offshore zone, the seafloor being too deep to correctly be represented within StrathE2E, a imaginary floating layer of sediment called “overhang” will be added to the model to represent deep sea interactions (Figure 2).
The perimeter of the model domain is defined by a 600m depth contour and the Norwegian coastline. Open ocean boundaries occur wherever there is no coastline. In consultation with local collaborators, we imposed additional boundaries to limit the eastward extent of the model domain at 42.2W and Southward along 33.75S, 54.35W to 36S, 50W. This coincides with changes in the pattern of fishing effort according to Global Fishing Watch, and Brazil’s exclusive economic zone.
In 2023 a Random Forest model was trained on a Norwegian Geological Survey sediment map to return a 1/100th degree resolution atlas of seabed sediment properties for the Barents Sea and North East Greenland (laverick2023?). This model was repurposed to predict for the Norwegian shelf to return area proportions of depth zones and seabed habitats for StrathE2E. The model provides gridded data sets of bathymetry, mean grain size, mud, sand and gravel content.
Porosity (proportion by volume of interstitial water) and permeability of each sediment habitat were derived from median grain sizes using empirically-based relationships.
\[log_{10}(porosity) = p_3 + p_4\left(\frac{1}{1+e^{(\frac{-log_{10}(D_{50})-p_1}{p_2})}}\right)\] D50 = median grain size (mm); parameters p1 = -1.227, p2 = -0.270, p3 = -0.436, p4 = 0.366 (Heath et al. 2015)
\[permeability = 10^{p_5}∙D_{50}^{*p_6}\]
where D50* = 0.11 ≤ D50 ≤ 0.50 p5 = -9.213, p6 = 4.615 (heath15?).
These relationships are coded into the StrathE2E2 R-package with the parameters in the csv setup file for the North Sea model. The parameters are probably a reasonable starting point for any future model of a new region. Derivation of the parameters is described in the following text sub-sections.
Values for each sediment type derived from parameterised relationships between total organic nitrogen content of sediments (TON%, percent by weight), mud content (mud%, percent by weight) and median grain size (D50, mm).
\[mud\% = 10^{p_7}∙𝐷_{50}^{𝑝_8}\] p7 = 0.657, p8 = -0.800
\[TON\% = 10^{𝑝_9}∙mud\%^{𝑝_{10}}\]
p9 = -1.965, p10 = 0.590
Proportion of TON estimated to be refractory = 0.9
These relationships are coded into the StrathE2E2 R-package with the parameters in the csv setup file for the North Sea model. The relationships and parameters are probably a reasonable starting point for any future model of a new region, though there are clear regional variations. Derivation of the parameters is described in the following sub-sections.
Table 1: Area-proportions of the inshore and offshore zones and the thicknesses of the water column layers. The sea surface area of the model domain is an estimated 771393.547989355 km2.
| Property | Inshore/shallow | Offshore/deep |
|---|---|---|
| Sea-surface area proportion | 0.1830 | 0.8170 |
| Upper layer thickness (m) | 261.0856 | 60.0000 |
| Lower layer thickness (m) | NA | 538.7608 |
We derived the area-proportions of seabed habitat in the inshore and offshore zones from output of the Random Forest model described by Laverick et al. (laverick2023?). The model allows for the prediction of a range of seabed data, including the presence of rock, the percentage of mud, sand and gravel fractions in the sediments and the whole–sediment mean grain size. These values are derived from habitat classes used by the Norwegian Geological survey in partnership with the Russian Federal State Unitarian Research and Production Company for Geological Sea Survey (NGU-SEVMORGEO). We assigned the NGU-SEVMORGEO sediment classes as fine, medium, coarse, or absence of sediment habitats within each zone (Figure 2). The actual area of each habitat was then the sum of the areas of each set of assigned cells (Table 2).
Table 2: Area proportions and other characteristics of the 8 seabed habitat classes defined in the model by depth, rock or sediment type. The sea surface area of the model domain is an estimated 771393.547989355 km2. Grain size is the median in mm, Permeability in units of m2, nitrogen content in %dw.
| Habitat | Sediment | Area Proportion | Grain size | Porosity | Permeability | Nitrogen content |
|---|---|---|---|---|---|---|
| Inshore - Shallow | ||||||
| S0 | None (Rock) | 0.0464 | NA | NA | NA | NA |
| S1 | Fine | 0.0337 | 0.0078 | 0.4429 | 0.0000 | 0.2700 |
| S2 | Medium | 0.0893 | 0.3536 | 0.3939 | 0.0000 | 0.0270 |
| S3 | Coarse | 0.0137 | 11.3137 | 0.3785 | 0.0000 | 0.0000 |
| Offshore - Deep | ||||||
| D0 | None (Overhang) | 0.8170 | NA | NA | NA | NA |
Log-transformed porosity has been shown to have a sigmoidal relationship with log10(median grain size) (D50, mm) (wilson18?):
\[log_{10}(porosity) = p_3 + p_4\left(\frac{1}{1+e^{(\frac{-log_{10}(D_{50})-p_1}{p_2})}}\right)\]
We use this relationship to calculate porosity for sea bed sediments in the Norwegian Sea (Table 2), using an alternative parameterisation to Wilson (pace21?). This alternative set of parameters extends the relationship to fine, muddy sediments (Table 3).
Table 3: The four parameters for the function relating sediment porosity to median grain size. From Pace et al. (in review)
| P1 | P2 | P3 | P4 |
|---|---|---|---|
| -1.035 | -0.314 | -0.435 | 0.302 |
Hydraulic conductivity (H, m.s-1) represents the ease with which fluids flow through the particle grain matrix. The related term ‘permeability’ (m-2) is a measure of the connectedness of the fluid filled void spaces between the particle grains. Permeability is a function only of the sediment matrix, whilst conductivity is a function of both the sediment and the permeating fluid, in particular the fluid viscosity and density. Hydraulic conductivity is related to permeability by:
\[H = Permeability∙fluid\;density∙\frac{𝑔}{dynamic\;viscosity}\]
where: seawater density = 1027 kg.m-3 at salinity 35 and temperature 10°C; seawater dynamic viscosity = 1.48 x 10-3 kg.m-1.s-1 at salinity 35 and temperature 10°C; g = acceleration due to gravity = 9.8 m.s-1
Hence, \(H = Permeability · 6.8004·10^6\) (m.s-1 at salinity 35 and temperature 10°C)
Whole sediment permeability can be related to the proportion of sediment classed as mud (D50 < 62 μm) (pace21?). This relationship was used in the production of the Norwegian Basin model (Table 2).
The magnitude of the static (refactory) organic nitrogen detritus pool in each sediment type is a required input to the model. The code includes an option to impute values from empirical relationships between total organic nitrogen (TON) and mud content, and between mud content and median grain size. This relationship has been documented in the North Sea implementation of the temperate StrathE2E2 package (Heath et al. 2021), and is based on sediment data off northeast Scotland.
Fixed parameters defining the proportion of ingested mass of food that contributes to new body tissue, after subtracting defecation and the metabolic costs of digestion and synthesis (Heath 2012).
Proportion of biomass lost to ammonia per day due to non-feeding related metabolism at a given reference temperature. Rates for individual guilds broadly related to typical body mass of representative species. Temperature dependency following a Q10 function.
Separate Q10 values for autotrophic uptake of nutrient, heterotrophic feeding, and heterotrophic metabolism based on literature data.
Light saturation intensity for nutrient uptake cannot be treated as a fitted value since it is confounded with other uptake parameters. Value estimated from survey of laboratory experiments.
Guild-level values derived by surveying the literature.
The living resource guilds in the model represent a mixture of harvestable and non-harvestable species, especially the invertebrate guilds. The density threshold parameter sets a limit for the guild biomass below which the harvestable species are assumed to be exhausted. Values set from analysis of trawl, plankton and benthos survey species biomass compositions.
The carnivorous zooplankton guild is a key component of the food web, predated on by all the fish and top-predators. However it represents an extremely diverse range of fauna many of which are not edible in significant quantities by the guild predators, e.g. scyphomedusae. A minimum edible threshold is set to ensure that the guild as a whole cannot be extirpated by predation. The value is a rough estimate of scyphomedusae biomass.
For the fish guilds the dates were obtained from literature survey. The annual weight-specific fecundity is assumed to be shed uniformly between the start and end dates of spawning. Pelagic planktivorous fish in the model are approximated as Norwegian Spring-Spawning herrings (Clupea harengus) as it is the major biomass in the region. The spawning dates were calculated from (Garcia et al. 2021) to be between February and March. Demersal fish spawning dates were approximated by the spawning dates of the cod (Gadus morhua), saithe (Pollachius virens) and haddock (Melanogrammus aeglefinus) from (Eikefjord 2023).
Obtained from literature survey. The annual cohort of larvae/juveniles of each fish and benthos guild is assumed to recruit to the settled stage at a uniform daily rate between the start and end dates. Again, the guild pelagic planktivorous fish in the model is approximated by Norwegian Spring-Spawning herring. The herring in larval stage, leaves the study area for the Barents Sea for 3-4 years and join the adult population at the end of their wintering (Holst, Røttingen, and Melle 2004). Demersal fish recruitment dates were also approximated by the recruitment dates of the cod (Gadus morhua), saithe (Pollachius virens) and haddock (Melanogrammus aeglefinus) from (Eikefjord 2023).
The annual weight-specific fecundity and recruitment for the benthos guilds are assumed to be uniform between the start and end dates. For the Scavenger and carnivorous benthos guild, the crab Cancer pagurus and the two sea urchins Echinus esculentus and Strongylocentrotus droebachiensis were elected to obtain timing dates (Bakke 2019; hamed2019?; James and Siikavuopio 2012). Finally, for the Suspension/deposit feeding benthos guild, it was particularly difficult to select reference species in the area and even more finding coherent spawning and recruitment dates. A large number of species have been reported in the area (Buhl-Mortensen, Hodnesdal, and Thorsnes 2015) going from cold water corals like Desmophyllum pertusum (formerly Lophelia pertusa ), to sponges and to shells. However, few specific dates for spawning were available and, when available, several species had multiple spawning per year which could not be included into the model. The Continuous Plankton Recorder (CPR) had only at this time, data from fish larvae survey so was not usable. We therefore decided to fix the spawning and recruitment dates to the dates of Desmophyllum pertusum (Brooke and Järnegren 2013; Järnegren and Kutti 2014).
Migratory fish in the Norwegian Sea model are assumed to be Atlantic mackerel and the Blue whiting. The spawning for Atlantic mackerel takes place off southwest Ireland in April and between February and April on the West of the British Isles for the Blue whiting. After spawning, both species rapidly migrate to summer feeding zones thousands of kilometers northwards along the continental shelf edge to the Norwegian and Barents Seas. (Iversen 2004; Bachiller et al. 2016). Some spawning of Atlantic mackerel have been observed in the Norwegian Sea feeding area (dos Santos Schmidt et al. 2024).
For the purposes of the model, we assume that there is no feedback between fishing and environmental conditions in the Norwegian Sea and the biomass and migration patterns of the whole northeast Atlantic mackerel or Blue whiting stock. In this version of StrathE2E2 the timing of immigration and emigration, and the mass influx across the ocean boundary during the annual immigration phase are treated as period-specific external driving data.
Data on the ‘global’ stock of northeast Atlantic mackerel and Blue whiting (wet biomass) are available from stock assessments (ICES 2020b), and converted to molar nitrogen mass using appropriate conversion ratios (Pedersen et al. 2021). The proportion of the migrating stock entering the Norwegian Sea (Nøttestad et al. 2016; Ekerhovd 2010), and the timing of the inward and outward migrations (Iversen 2002; Payne et al. 2012) are estimated from the literature. A residual proportion of the peak abundance in the North Sea remaining as residents (if any) is estimated from summer trawl survey data. The model setup code calculates the parameters which are needed in the ecology model.These are the only fixed (i.e. non-fitted) ecology model parameters which are period-specific.
Table 4: Biological event timing parameters, constant across the 2011-2070 time period. The data are processed in the model setup to calculate the immigration flux parameters needed in the ecology model.
| Parameter | Value |
|---|---|
| Pelagic fish spawning start day | 58 |
| Pelagic fish spawning duration (days) | 21 |
| Pelagic fish recruitment start day | 255 |
| Pelagic fish recruitment duration (days) | 150 |
| Demersal fish spawning start day | 43 |
| Demersal fish spawning duration (days) | 45 |
| Demersal fish recruitment start day | 95 |
| Demersal fish recruitment duration (days) | 30 |
| Filt/dep benthos spawning start day | 20 |
| Filt/dep benthos spawning duration (days) | 50 |
| Filt/dep benthos recruitment start day | 56 |
| Filt/dep benthos recruitment duration (days) | 56 |
| Carn/scav benthos spawning start day | 10 |
Table 4 Continued.
| Parameter | Value |
|---|---|
| Carn/scav benthos spawning duration (days) | 110.00 |
| Carn/scav benthos recruitment start day | 240.00 |
| Carn/scav benthos recruitment duration (days) | 90.00 |
| Migratory fish switch (0=off 1=on) | 1.00 |
| Migratory fish ocean biomass (Tonnes wet weight) | 11230000.00 |
| Migratory fish carbon to wet weight (g/g) | 0.12 |
| Model domain sea surface area (km2) | 771393.55 |
| Propn of ocean population entering model domain each year | 0.45 |
| Imigration start day | 91.00 |
| Imigration end day (must be later than start day even if migration disabled) | 270.00 |
| Propn of peak popn in model domain which remains and does not emigrate | 0.20 |
| Emigration start day | 300.00 |
| Emigration end day (must be later than start day even if migration disabled) | 360.00 |
Monthly resolution time-varying physical and chemical driving parameters for the model were derived from a variety of sources:
Details of how these data were processed are given below, supported by the nemomedusR and MiMeMo.tools packages.
Four different NEMO-ERSEM runs were used to parameterise different versions of the Norwegian Sea implementation of StrathE2E. These four runs are a 2x2 factorial cross of two future projected scenarios (SSP126 and SSP370) with a historical hindcast from 2015, and forcing by two atmospheric models (CNRM and GFDL). In the following sections model output was processed for a 2010-2019 baseline period, and then decadal projections from 2020-2029 until 2060-2069.
Vertical diffusivity from the NEMO-ERSEM coupled hydro-geochemical model output was interpolated for each grid cell at the 60m boundary depth between the shallow and deep layers of the offshore zone, and the 600m boundary at the deep sea overhang. These values were summarised as monthly averages into period-specific climatological annual cycles of data for decadal periods for all combinations of SSPs and forcings.
Derived by monthly averaging values at grid points within the inshore and vertical layers of the offshore zones from the NEMO-ERSEM coupled hydro-geochemical model output, weighted by grid point volumes. Values were summarised into period-specific climatological annual cycles of data.
Monthly averaged values of inorganic suspended particulate matter in sea water are available from the Globcolour project, starting from September 1997. These data are derived from satellite observations using the algorithm of Gohin (gohin11?). Data were downloaded from the ftp server (ftp://ftp.hermes.acri.fr/GLOB/merged/month/). We summarised these values as zonal statistics for the model domain to acquire a climatological annual cycle of data for the 2010-2019 simulation period only.
Light attenuation in open water was parameterised from a linear relationship between the light attenuation coefficient and suspended particulate matter concentration (SPM) (Devlin et al., 2008).
Derived from HadGEM2-ES model output (jones11?) which forces the NEMO-ERSEM model run used throughout our implementation. Monthly mean values were summarised into climatological annual cycles of data for decadal periods and NEMO-ERSEM ensembles.
Sourced from the “histsoc” files for a 1901 - 2021 hindcast as monthly averages (yang22?), available from CDS. Monthly values were summarised into climatological annual cycles of monthly oxidised and reduced nitrogen deposition rates extracted for 2010-2019. ISIMIPb Projections for SSP126 and SSP370 were processed in the same way for decadal periods up to and including the 2060s.
Freshwater inflow derived from HadGEM2-ES model output (jones11?) which forces the NEMO-MEDUSA model run used throughout our implementation. Monthly values were summaries into a climatological annual cycle of data for both the 2010-2019 and 2040-2049 periods.
Monthly averaged daily inflow and outflow volume fluxes derived by integrating daily mean velocities directed perpendicular to transects along the model domain boundary at grid points in each depth layer along transects through outputs from the NEMO-ERSEM coupled hydro-geochemical model output. Monthly averaged daily inflow volume fluxes then divided by the volume of the receiving layer in the model domain to estimate a daily flushing rate.
NEMO-ERSEM outputs included nitrate, ammonia, phytoplankton and suspended detritus. We calculated the depth-averaged concentrations for pixels within the shallow and deep layers of StrathE2E. We then sampled the pixels using the same transects around the model domain as for sampling volume fluxes. Only transects where water flowed into the model domain were sampled, and the average concentration of inflowing waters for target variables was calculated weighting by the flow rate across a transect and the cross-sectional area represented by a transect (average depth and length). Concentrations were then averaged into climatological annual cycles. The flow weighted averaging of concnetrations was calculated on the 5-day NEMO-ERSEM timestep, before averaging to a monthly climatology of concentrations.
The key configuration data for the fishing fleet model are the definitions of the gears in terms of their power with respect to each of the harvestable resource guilds, discarding rates, processing-at-sea rates, and their seabed abrasion rates. These can be regarded as static parameters for each gear.
An additional class of static parameters is the scaling coefficients between effort (activity x power) and the harvest ratio generated on each model resource guild. These parameters have to be derived by fitting.
Finally, there are parameters which we can consider as driving data since they would be expected to vary with time. These are the activity rates of each gear, and their spatial distributions across the habitat types.
Our principal data sources for the Norwegian Sea were were ICES (https://www.ices.dk/data/dataset-collections/Pages/Fish-catch-and-stock-assessment.aspx), IMR , the Norwegian directorate of fisheries (Fiskeridirektoratet) (https://www.fiskeridir.no/Tall-og-analyse/AApne-data), and Global Fishing Watch (kroosdama2018?). These were supplemented with additional data sources to improve the representation of catch, bycatch, discard, and small recreational and artisanal fisheries.
Static parameters for the fishing fleet model were taken from the North Sea implementation (Heath et al. 2021), with the new set of gears operating in the Barents Sea reconciled with StrathE2E gear types as detailed below (Table 5). These parameters would be expected to remain constant over time, so any changes invoked would imply a change in the design or operation of a gear type.
Table 5: The gear labelling systems of the Norwegian, Faroese and Icelandic Directorates of Fisheries, IMR and Global Fishing Watch, were reconciled with StrathE2E gear types. Gears were condensed considering their target species and their likely impact on the sea-bed.
| StrathE2E Gear | Gear definition | Global Fishing Watch gears | IMR gear codes | Icelandic directorate gears | Faroese directorate gears | IMR reference fleet gears |
|---|---|---|---|---|---|---|
| Dredging | Dredge | dredge_fishing | 81 | |||
| Dredging | dredge_fishing | 81 | Shellfish dredge | |||
| Dredging | dredge_fishing | 81 | Scallop dredge | Scallop dredge | ||
| Dropped | Other | 80 | ||||
| Dropped | Unkown | NK | ||||
| Harpoons | Harpoon and similar unspecified types | 70 | ||||
| Kelp harvesting | Kelp harvesting | Seaweed | ||||
| Longlines and Gillnets | Undefined hook gear | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 30 | |||
| Longlines and Gillnets | Other hook and line | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 32 | Long line: hand line | line | |
| Longlines and Gillnets | Floating hooks | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 31 | Handline | Long line: drifting | line |
| Longlines and Gillnets | Jigging | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 33 | Jigging line: automatic | line | |
| Longlines and Gillnets | Jigging | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 33 | Jigging line: manual | line | |
| Longlines and Gillnets | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | Bottom longline | Long line: demersal | line | ||
| Longlines and Gillnets | Undefined net | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 20 | line | ||
| Longlines and Gillnets | Gillnet (static) | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 22 | Bottom gillnet | Gill net: demersal | |
| Longlines and Gillnets | Gillnet (drifting) | pole_and_line+set_longlines+squid_jigger+drifting_longlines+set_gillnets | 21 | Gill net: drifting | ||
| Pelagic Trawlers | Mid-water otter trawl | Pelagic_trawlers | OTM | Pelagic trawl: otter trawl | ||
| Pelagic Trawlers | Pelagic trawl | Pelagic_trawlers | 53 | Pelagic trawl | ||
| Pelagic Trawlers | Pelagic_trawlers | 53 | Pelagic trawl: pair trawl | |||
| Pelagic Trawlers | Pelagic trawl (pair) | Pelagic_trawlers | 54 | |||
| Pots | Pot | pots_and_traps | 42 | |||
| Pots | Pots | pots_and_traps | FPO | |||
| Pots | pots_and_traps | 41 | Whelk trap | Traps | ||
| Recreational | Recreational | DUMMY | ||||
| Seiners | Purse seine | Seiners | 11 | Purse seine | ||
| Seiners | Undefined seine net | Seiners | 10 | |||
| Seiners | Danish seine | Seiners | 61 | Danish seine | Danish seine | |
| Seiners | Seiners | Seine: single ship | ||||
| Seiners | Seiners | Seine: ring net | ||||
| Seiners | Seiners | Seine: two ships | ||||
| Shelf Trawlers | Undefined trawl | Shelf_trawlers | 50 | trawl | ||
| Shelf Trawlers | Double trawl | Shelf_trawlers | 58 | Bottom trawl: double trawl | trawl | |
| Shelf Trawlers | Triple trawl | Shelf_trawlers | 59 | trawl | ||
| Shelf Trawlers | Beam trawl | Shelf_trawlers | 56 | trawl | ||
| Shelf Trawlers | Bottom trawl | Shelf_trawlers | 51 | Bottom trawl | trawl | |
| Shelf Trawlers | Bottom otter trawl | Shelf_trawlers | OTB | trawl | ||
| Shelf Trawlers | Bottom pair trawl | Shelf_trawlers | PTB | Bottom trawl: pair trawling | trawl | |
| Shelf Trawlers | Mid-water pair trawl | Shelf_trawlers | PTM | trawl | ||
| Shelf Trawlers | Otter twin trawls | Shelf_trawlers | OTT | Bottom trawl: otter trawl | ||
| Shelf Trawlers | Shrimp trawl | Shelf_trawlers | 55 | Shrimp trawl | Bottom trawl: prawn | trawl |
| Shelf Trawlers | Shelf_trawlers | Neophrops trawl | trawl |
(Nedreaas, Borge, and Godøy 2006; T. Clegg and Williams 2020)#discards
The following briefly describes the potentially time-varying driving data for the fishing fleet model.
An annual average was calculated for the 2010-2019 period using the data available within this time period for the Norwegian fishing fleet (Norwegian Directorate of Fisheries), the EU fleet (STECF), the Icelandic Fishing fleet (Icelandic Directorate of Fisheries) and the Faroese fishing fleet (Statistics Faroe Islands). Values were summed and inflated for estimates of total international activity using ICES data to infer the missing International (not Norway, Iceland, Faroe or EU countries) catch in the Norwegian Sea. This assumes the International fishing fleet operates a similar gear distribution to the EU, Norwegian, Icelandic and Faroese fleets when combined.
An annual average was calculated for the 2010-2019 period using the data available within this time period for the Norwegian fishing fleet(IMR) and the EU fleet (STECF). Values were summed and inflated for estimates of total international activity using Global fishing watch (GFW) data to infer the missing International (not Norway or EU countries) activity in the Norwegian Sea for the GFW gears. This assumes the International fishing fleet operates a similar gear distribution (within static and mobile gear types) to the EU and Norwegian fleets when combined.
Proportion of domain-wide annual average activity rate over each seabed habitat type, derived by overlaying spatial distributions of activity from IMR (Norwegian), STECF (EU), and Global Fishing Watch (International), onto spatial distributions of seabed sediment types derived from the atlas of sediment properties (laverick2023?).
IMR provided us with daily catch and activity data for the Norwegian fishing fleet on request. This data was broken down by species caught and gear used in fishing areas. We limited the data to 2010-2019, from the first year of the electronic reporting system to the last complete year of data. Cetacean records appeared to start from 2013, so averages were calculated for cetaceans from 2013-2019.
We aggregated gears and species to StrathE2E gear types and guilds. Data was totaled within years, then averaged across the target time period. Effort and landings were summed by gear and guild within years and IMR area codes.
IMR areas do not perfectly align with the StrathE2E model domain, we therefore applied a correction factor to landings and effort to account for IMR data falling outside the model domain. We summarised the data available from global fishing watch from 2010-2019 into average annual 0.01° grids of total yearly fishing activity for each of GFW guilds (Table 5). We intersected the polygons representing IMR area codes, and the StathE2E model domain, and calculated the total gear activity within each polygon according to GFW. We then calculated the proportion of GFW gear activity for each IMR area code which fell within the StrathE2E model domain.
We used the same approach to calculate the proportion
Arctic Fisheries Working Group (AFWG) reports since 2010 include catches of cod by non-commercial fishing activity in assessments for the coastal cod stock which inhabits the inshore waters of western and northern Norway. Coastal cod are considered to be distinct from the more abundant Northeast Arctic cod which are mostly found offshore but migrate inshore to breed in the spring. The noncommercial activity includes both fishing-tourism (businesses offering fishing trips to visitors), and recreational fishing by Norwegian residents. The former is a rapidly growing sector in Norway, but still smaller in terms of catch than the recreational sector. In some years the combined tourist and recreational catch was estimated to account for up to one-third of the total catch of coastal cod. According to (ICES 2020a), since 2010 seven thousand tonnes of the Norwegian cod quota has been set aside annually to cover the catches taken in the recreational and tourist fisheries and to motivate young people to become fishers.
The fishing methods for non-commercial activity are primarily hooks for the tourist sector, and hooks and gillnets for the recreational sector. Many tourist businesses operate a catch and release system. Fish which are released are assumed to survive and are not included in assessments of tourist catch (Vølstad et al. 2011). Likewise, some recreational fishers offer a proportion their catches for sale. ICES assessments assume that these sales are already included in the catch statistics compiled by the Norwegian Directorate for Fisheries, and hence not classified as part of the recreational catch.
Subject to the above limitations, the 2010 AFWG prepared a record of both recreational and tourist catches of coastal cod between 1984 and 2009, based on studies by (Hallenstvedt 2004; Anon 2005), and annual surveys on the number of Norwegian residents who said they had been fishing in the sea. These strands of evidence indicated the total recreational and tourist catch of cod to be 13,400 tonnes in 2004 and the tourist catch 1,100 tonnes. It was estimated that participation in sea fishing tourism increased by 19% per year between 1995 and 2000, by 16% per year until 2004, and then by 10% per year up to 2009 (ICES 2010). No new data have become available since 2009 and subsequent AFWGs up to 2020 have simply projected the 2009 estimated total catch of 12,700 tonnes forwards year on year, without distinguishing between the tourist and recreation sectors (ICES 2020a). Here, we have conservatively assumed that tourism catches have increased by 5% year-on-year since 2009, and derived the recreational catch as the difference between the stated total (ICES 2020a) and our estimated tourist catch (Table 6).
Table 6 provides the total annual recreational and tourist catches of coastal cod, which occurs in western and northern Norway, northward of around 62N. However, we required the catch not just of cod but also of the other species taken by these fisheries, and the subset of these that originate from our Norwegian Sea model area. Data from (Vølstad et al. 2011) provide some information to attempt this extrapolation.
(Vølstad et al. 2011) surveyed Norwegian sea fishing tourism businesses in 2009 3)to gather data on the level of activity (boat days), participants and catch composition. Catch quantities by species were integrated for regions north and south of 62N (Table 7). There was a clear different in composition, with cod forming over 50% of the catch north of 62N, and less than 10% to the south. By way of corroboration, ICES (2010) also estimated that cod formed around 50% of the recreational and tourism catch north of 62N during the early 2000’s. South of 62N, saithe was the main species caught, followed by mackerel in 2009 (Vølstad et al. 2011). Apart from mackerel, all the species were members of the demersal fish guild in our model.
In the absence of any other catch composition data we estimated the total demersal catch, and the mackerel catch for the area north of 62N in each year from the 2009 data, assuming a constant ratio of all demersal species : cod of 1.857, and mackerel : cod of 0.0086 (Table 7).
Tourism and recreation catches along the Norwegian coast bordering our Barents Sea model would clearly be less than the total north of 62N. However, the breakdown of catches in 2009 at finer resolution than that reported by Vølstad et al. (2011) was not available. Hence we crudely estimated the proportion of annual catches from north of 62N that might have been taken in our model area based on the proportion of sea fishing tourism businesses participating in the 2009 study which were located in the Troms and Finmark areas (Figure 3; 11 out of 52 businesses north of 62N; 21.57%). The results are estimates, albeit obviously crude, of the catch of demersal fish and migratory fish (mackerel) guilds in our Barents Sea model domain (Table 8).
Table 6: Coastal cod catch from ICES areas 1 and 2 by tourist businesses recreational fishers. Between 1984 and 2009 the tourist only catch is from ICES ((2010); Table 2.1c), while the combined tourist + recreation catch (1984-2019) is from ICES ((2020); Table 2.1e). Between 2010 ad 2019 the tourist catch (grey shaded) is assumed to increase at 5% per year (ICES 2020). The recreation-only catch is the difference between the total and tourism
| Year | Combined tourism & recreation catch (tonnes) | Tourist catch (tonnes) | Implied recreation catch (tonnes) = total- tourist |
|---|---|---|---|
| 1984 | 13300 | 0 | 13300 |
| 1985 | 13400 | 0 | 13400 |
| 1986 | 13500 | 0 | 13500 |
| 1987 | 13500 | 0 | 13500 |
| 1988 | 13600 | 0 | 13600 |
| 1989 | 13700 | 100 | 13600 |
| 1990 | 14500 | 100 | 14400 |
| 1991 | 15300 | 100 | 15200 |
| 1992 | 16100 | 100 | 16000 |
| 1993 | 14800 | 100 | 14700 |
| 1994 | 14700 | 100 | 14600 |
| 1995 | 14700 | 200 | 14500 |
| 1996 | 14500 | 200 | 14300 |
| 1997 | 14500 | 300 | 14200 |
| 1998 | 14600 | 300 | 14300 |
| 1999 | 13900 | 400 | 13500 |
| 2000 | 13600 | 500 | 13100 |
| 2001 | 13400 | 700 | 12700 |
| 2002 | 13600 | 800 | 12800 |
| 2003 | 13900 | 900 | 13000 |
| 2004 | 13400 | 1100 | 12300 |
| 2005 | 13200 | 1200 | 12000 |
| 2006 | 13000 | 1300 | 11700 |
| 2007 | 13000 | 1500 | 11500 |
| 2008 | 12800 | 1600 | 11200 |
| 2009 | 12700 | 1800 | 10900 |
| 2010 | 12700 | 1890 | 10810 |
| 2011 | 12700 | 1985 | 10716 |
| 2012 | 12700 | 2084 | 10616 |
| 2013 | 12700 | 2188 | 10512 |
| 2014 | 12700 | 2297 | 10403 |
| 2015 | 12700 | 2412 | 10288 |
| 2016 | 12700 | 2533 | 10167 |
| 2017 | 12700 | 2659 | 10041 |
| 2018 | 12700 | 2792 | 9908 |
| 2019 | 12700 | 2932 | 9768 |
| Mean 2011-2019 | 12700 | 2431 | 10269 |
Table 7: Data for the 2009 national survey of tourist recreational fishing in Norway from (Vølstad et al. 2011), Table 2. The data were presented for two regions, North and South of 62N. The additional column shown here for the Finmark and Troms area is a 21.57% subset of the data for “North of 62N” based the proportion (11 out of 52) of the tourist businesses in “North of 62N” which were located in these administrative areas. Values are the annual catch weight in tonnes.
| Measure | North of 62N | South of 62N | Estimated Finmark and Troms regions (21.57% of “North of 62N”) | Other regions (78.43% of “North of 62N”) | ...6 | 0.65384615384615385 |
|---|---|---|---|---|---|---|
| Cod | 1586.00000 | 27.000 | 335.50000 | 1.243900e+03 | NA | NA |
| Haddock | 115.10000 | 9.400 | 24.30000 | 9.027293e+01 | NA | NA |
| Saithe | 825.20000 | 208.000 | 174.60000 | 6.472044e+02 | NA | NA |
| Pollack | 81.10000 | 21.400 | 17.20000 | 6.360673e+01 | NA | NA |
| Hallibut | 79.70000 | 0.200 | 16.90000 | 6.250871e+01 | NA | NA |
| Mackerel | 13.60000 | 54.400 | 2.90000 | 1.066648e+01 | NA | NA |
| Ling | 68.50000 | 40.400 | 14.50000 | 5.372455e+01 | NA | NA |
| Tusk | 173.70000 | 15.900 | 36.70000 | 1.362329e+02 | NA | NA |
| Wolffish | 15.30000 | 0.300 | 3.20000 | 1.199979e+01 | NA | NA |
| Total weight | 2958.20000 | 377.000 | 625.80000 | 2.320116e+03 | NA | NA |
| Total weight excl. mackerel | 2944.60000 | 322.600 | 622.90000 | 2.309450e+03 | NA | NA |
| Ratio of total demersal guild : cod | 1.85700 | 11.948 | 1.85700 | 1.856620e+00 | NA | NA |
| Ratio of mackerel : cod | 0.00858 | 2.015 | 0.00858 | 8.575032e-03 | NA | NA |
Table 8: Data for the 2009 national survey of tourist recreational fishing in Norway from (Vølstad et al. 2011), Table 2. The data were presented for two regions, North and South of 62N. The additional column shown here for the Finmark and Troms area is a 21.57% subset of the data for “North of 62N” based the proportion (11 out of 52) of the tourist businesses in “North of 62N” which were located in these administrative areas. Values are the annual catch weight in tonnes.
| Year | Tourist.Cod | Tourist.Total Demersal | Tourist.Mackerel | Recreation.Cod | Recreation.Total Demersal | Recreation.Mackerel | Combined.Total Demersal | Combined.Mackerel |
|---|---|---|---|---|---|---|---|---|
| 1984 | 0 | 0 | 0 | 2479 | 4603 | 21 | 4603 | 21 |
| 1985 | 0 | 0 | 0 | 2498 | 4638 | 21 | 4638 | 21 |
| 1986 | 0 | 0 | 0 | 2516 | 4673 | 22 | 4673 | 22 |
| 1987 | 0 | 0 | 0 | 2516 | 4673 | 22 | 4673 | 22 |
| 1988 | 0 | 0 | 0 | 2535 | 4707 | 22 | 4707 | 22 |
| 1989 | 19 | 35 | 0 | 2535 | 4707 | 22 | 4742 | 22 |
| 1990 | 19 | 35 | 0 | 2684 | 4984 | 23 | 5019 | 23 |
| 1991 | 19 | 35 | 0 | 2833 | 5261 | 24 | 5296 | 24 |
| 1992 | 19 | 35 | 0 | 2982 | 5538 | 26 | 5573 | 26 |
| 1993 | 19 | 35 | 0 | 2740 | 5088 | 23 | 5123 | 24 |
| 1994 | 19 | 35 | 0 | 2721 | 5053 | 23 | 5088 | 23 |
| 1995 | 37 | 69 | 0 | 2703 | 5019 | 23 | 5088 | 23 |
| 1996 | 37 | 69 | 0 | 2665 | 4950 | 23 | 5019 | 23 |
| 1997 | 56 | 104 | 0 | 2647 | 4915 | 23 | 5019 | 23 |
| 1998 | 56 | 104 | 0 | 2665 | 4950 | 23 | 5053 | 23 |
| 1999 | 75 | 138 | 1 | 2516 | 4673 | 22 | 4811 | 22 |
| 2000 | 93 | 173 | 1 | 2442 | 4534 | 21 | 4707 | 22 |
| 2001 | 130 | 242 | 1 | 2367 | 4396 | 20 | 4638 | 21 |
| 2002 | 149 | 277 | 1 | 2386 | 4430 | 20 | 4707 | 22 |
| 2003 | 168 | 312 | 1 | 2423 | 4500 | 21 | 4811 | 22 |
| 2004 | 205 | 381 | 2 | 2293 | 4257 | 20 | 4638 | 21 |
| 2005 | 224 | 415 | 2 | 2237 | 4153 | 19 | 4569 | 21 |
| 2006 | 242 | 450 | 2 | 2181 | 4050 | 19 | 4500 | 21 |
| 2007 | 280 | 519 | 2 | 2143 | 3980 | 18 | 4500 | 21 |
| 2008 | 298 | 554 | 3 | 2088 | 3877 | 18 | 4430 | 20 |
| 2009 | 336 | 623 | 3 | 2032 | 3773 | 17 | 4396 | 20 |
| 2010 | 352 | 654 | 3 | 2015 | 3742 | 17 | 4396 | 20 |
| 2011 | 370 | 687 | 3 | 1997 | 3709 | 17 | 4396 | 20 |
| 2012 | 388 | 721 | 3 | 1979 | 3675 | 17 | 4396 | 20 |
| 2013 | 408 | 757 | 3 | 1959 | 3638 | 17 | 4396 | 20 |
| 2014 | 428 | 795 | 4 | 1939 | 3601 | 17 | 4396 | 20 |
| 2015 | 450 | 835 | 4 | 1918 | 3561 | 16 | 4396 | 20 |
| 2016 | 472 | 877 | 4 | 1895 | 3519 | 16 | 4396 | 20 |
| 2017 | 496 | 920 | 4 | 1871 | 3475 | 16 | 4396 | 20 |
| 2018 | 520 | 967 | 4 | 1847 | 3429 | 16 | 4396 | 20 |
| 2019 | 546 | 1015 | 5 | 1821 | 3381 | 16 | 4396 | 20 |
| Mean 2011-2019 | 453 | 842 | 4 | 1914 | 3554 | 16 | 4396 | 20 |
The European commission’s Scientific, Technical, and Economic Committee for Fisheries (STECF) provide spatially explicit datasets on annual fish catch (https://stecf.jrc.ec.europa.eu/dd/fdi/spatial-landmap) and fishing effort (https://stecf.jrc.ec.europa.eu/dd/fdi/spatialeff- map) by the fishing fleets of member states around the world. Data is available for the years 2015-2018 by fishing gear. The landings dataset is further resolved by species caught. The data processing steps were similar to those described above for the Norwegian fishing fleet. 1. Once again we corrected landings and fishing effort according to the overlap between STECF reporting areas and our model domain. We summarised the data available from global fishing watch from 2012-2016 into average annual 0.01° grids of total yearly fishing activity for mobile and static gears. We intersected the polygons representing STECF reporting areaa, and the StathE2E model domain, and calculated the total mobile and static gear activity within each polygon according to GFW. We then calculated the proportion of mobile and static gear activity for each STECF area code which fell within the StrathE2E model domain and scaled the values reported by STECF. 2. We aggregated gears and species to StrathE2E gear types and guilds. Data was totaled within years, then averaged across the target time period. Effort and landings were summed by gear and guild within years. 3. We used the same approach to calculate the proportion of fishing effort across the 8 StrathE2E habitat types. Instead of the intersection between the StrathE2E model domain and the STECF reporting regions, we intersected the Norwegian Sea habitats with STECF reporting regions (Figure 2)). We then calculated the proportion of Global Fishing Watch gear activity for each reporting region which fell within the StrathE2E habitats. 4. Corrected landings were summed spatially and saved as a matrix by gear and guild. Corrected effort was totaled across area codes and saved by StrathE2E gear types. Corrected effort was also totaled across area codes and saved as a matrix by gears and habitats.
Though we have spatially explicit data of fisheries landings and effort for the Norwegian and EU fishing fleet in Norwegian Sea, a sizable portion of activity is missing. The Russian, Icelandic and Faroese and other international fishing fleet that we will call “International fishing fleet” have a notable presence in the Norwegian Sea according to ICES (ICES 2022). Unfortunately the ICES data does not resolve landings by gear used, and the spatial resolution of landings is coarse.
To approximate international catch we use the ICES data on fisheries landings from our internal database to calculate an inflation factor. 1. We divide the total weight landed per guild in area 27.2.a by the total international catch in the same area. 2. We sum the Norwegian and EU matrices described above, and multiply by the ICES correction factor per guild. This assumes the International fishing fleet has a distribution of fishing gears similar to the combined EU and Norwegian fishing fleet. 3. Kelp harvesting, Dredge fishing, and the tourist and recreational catch was added to the international matrix after the application of the inflation factor.
To approximate international catch we use the GFW data (kroosdama2018?) to calculate an inflation factor. 1. We divide the total effort by mobile and static gears by the total international effort represented by flags other than Russia. The appropriate correction factors were matched to StrathE2E gear types based on wthe GFW gears. The following gears were not inflated as they are unique to the Norwegian fleet: Harpoons, Dredge, Kelp harvesting, Recreational. 2. We sum the Norwegian and EU effort vectors described above, and multiply by the GFW correction factor per gear. 3. The tourist, recreational (described above), harpoons, dredge, kelp harvesting effort were added to the international effort after the application of the inflation factor. 4. Annual hours of fishing effort were then converted into to daily effort in s/m2.
Discard rates were calculated as the ratio of discarded weight to caught weight for each gear and guild.
In Norway, a discard ban was introduced in 1987 for some species and was gradually expanded to include all species by 2009. However, we believe the discard rate is not 0 for the Norwegian fleet. No data is available on the discards of the Norwegian fleet for every species and gears so we worked only with the available data from various sources. We got estimations from the longlines and trawls fisheries of vessels >28m using unpublished data from IMR (Huitfeldt Aspelund, Clegg, and Nedreaas 2024) in the ICES subregion 27.2.a (ADD FIGURE), calculated using the method described in (T. L. Clegg, Fuglebakk, and Ono 2023). We used a set data from IMR containing data on discards of cod, haddock and anglerfish in Norwegian coastal fisheries between 2012 and 2018 and data of discards of all species in coastal gillnets fisheries in 2018 (Hilde Sofie Fantoft and Nedreaas 2021). We estimated the bycatch of cetaceans from gillnets fisheries using data for the harbour porpoise and white-sided dolphin, assuming these species represent the vast majority of the bycatch of cetaceans (Moan et al. 2020). For the pinnipeds, we used discards data from gillnets fisheries from (Moan and Bjørge 2021). We We calculated the discard rates as \(discards/(landings+discards)\) and summarised to mean discard rates for StrathE2E gear classes and guilds.
Unrepresented combinations of gear and guild were automatically assigned a discard rate of 0. We manually set the discard rates for a number of guilds which are only targetted by a single specialist gear to 1. These were cetaceans, pinnipeds, birds, and macrophytes. Harpoons for cetaceans, Rifles for pinnipeds, and kelp harvesting for macrophytes were then all assigned a discard rate of 0.
Total catch by gear and guild was calculated by inflating landings according to EU discard rates before adding additional known discarded weight (described above for cetaceans, birds, planktivores and demersal fish).
Discarded weight was calculated as catch - landings.
Demersal non quota and quota limited were combined into a single Demersal guild for catch, landings and discards.
New discard rates reflecting all data sources were calculated as discarded weight / caught weight. When catch was 0 discard rates were set to 1 except for kelp harvesters which were assigned a discard rate of 0 for macrophytes.
Fishing power was calculated as catch / activity per gear.
Financial support for the development of the Norwegian Sea implementation of StrathE2E2 came from the European Union’s Horizon 2020 research and innovation programme (Mission Atlantic - No. 862428). We are grateful to: